10 July 2020

The Extended Heart Sutra: Sources

Very little critical or comparative work has been done on the extended Heart Sutra. I've tended to ignore it because there has been so much to do on the standard text. I've been interested in the origins of the text and the extended version is a later development. With most of my philological and historical work on the standard completed and awaiting publication, I have been making some notes about the extended text.

The majority of the extant Heart Sutra manuscripts and inscriptions are of the extended text. There are five canonical Chinese versions and two canonical Tibetan Versions. There is a Sanskrit manuscript in Japan and about a dozen manuscripts from Nepal have been described, although I think dozens more are extant. And then there are the Dunhuang texts—over 180 in Chinese and 90 in Tibetan—which reflect nine different version of the standard and extended texts and some hybrids. 

Most of the Extended texts conform to a single recension, i.e. they all seem to have common origins, though with many minor variations. I will call this Recension One (R1). There seem to be a number of variants of R1 and part of the purpose of the coming series of essays is to identify these through close comparison of the available documents. There is also a Recension Two (R2) which is represented by a single text, T 252 a canonical translation attributed to Fǎyuè 法月 (Skt. *Dharmacandra) and Lìyán 利言, and dated ca. 741 CE. If the attribution and date are reliable then this is the earliest evidence of the extended text.

In this post I'll describe the extant texts in more detail. In my last post, I raised the question of whether the Heart Sutra ever went to India. I concluded that we didn't have enough information but that it was plausible that the transmission was from China to Tibet and that Vimalamitra, as the man credited with the canonical Tibetan translation, might have had something to do with this. Although his connection to China seems tenuous, a private message from Joel Gruber (whose PhD dissertation was on Vimalamitra) suggests that he thinks it likely that Vimalamitra did visit China on his way to Tibet.

I will look at the Tibetan versions with the help of Jonathan Silk's critical edition. But I don't think we can say much about the Heart Sutra in other countries on the periphery of Tang China such as Khotan, Kucha, Korea, Japan, Mongolia, or Vietnam. This may be bias because I don't read any of the relevant languages, but my impression is that no textual innovations occurred in these countries. They adopted the standard Chinese text without modification, except perhaps chanting it in the local pronunciation - thereby ensuring that it was incomprehensible in either language. I'm open to being proved wrong about this, but in any case for now I'm covering Sanskrit, Chinese, and Tibetan, focussing on the early 8th to the late 9th Centuries. The scholarly (as opposed to the religious) history of the text in these other countries has yet to be written in English.



We now know that the Sanskrit text of the standard Heart Sutra was composed in China. We don't yet know where the extended text was composed.

The main exemplar of the Sanskrit, the manuscript from Hase Temple (Jap. Hasedera 長谷寺 aka Haseji), is said to have been brought from China in the early 9th Century. The manuscript also includes a Sanskrit text, a Chinese transliteration of the Sanskrit, and a copy of the Chinese translation by Prajñā (T 253). Müller's 1884 diplomatic edition was based on hand copies of the Hasedera manuscript. Vaidya's edition is based on it or on Müller's edition. Conze relies heavily on Müller.

It's not mentioned on the Hasedera website and it does not seem to have been digitised. It's unclear  to me where this document is now. 

Perhaps a few dozen Sanskrit manuscripts survive from Nepal, largely in European collections and largely undocumented. For example, the catalogue of the Nepalese-German Manuscript Cataloguing Project has many likely candidates that have not been fully described. However, these 19th and 20th Century copies are of little use in studying the text. In 2014, I described a manuscript from Nepal digitised as part of the Endangered Archives Program. Manuscript EAP676/2/5 required 142 footnotes to detail all the omissions, additions, and errors. These manuscripts are the result of many generations of copying without error checking because the scribes did not know Sanskrit. The  resulting documents are objects of magic and worship, not intended to be read. The effort required to describe each in detail is probably not worth it.

It seems that the standard and extended texts in Sanskrit may have been made from a different ancestor text (unrelated to T 252). The standard Sanskrit texts all contain extra-negations that make no sense, e.g. na prāptiḥ → na prāptir na aprātiḥ and in the nidānas: na avidyā  na vidyā na avidyā. These additions are present in the Hōryūji (法隆寺) manuscript and the Amoghavajra Chinese transliteration (T 256), our principal sources for the Sanskrit standard text.  Of the extended texts, these extra negations are found in the Hasedera ms. but not in any of the Chinese translations or the Nepalese manuscripts.

Sanskrit users were to be found in small numbers outside of India in antiquity, mainly working as translators.

Tibetan Texts

The one place that the extended text was definitely in use was Tibet because there it became the Heart Sutra. The standard text was largely overlooked although copies of the standard text transliterated in Tibetan script and translated into Tibetan language are amongst the cache of Heart Sutra texts from Dunhuang. 

Jonathan Silk (1994) has described two canonical recensions of the extended text in Tibet (TibA and TibB) in detail. TibA is in the tantra section of the Kanjur, while TibB is in the Prajñāpāramitā section. There is considerable variation and hybridisation between the two texts in the fourteen extant editions of the Kanjur, most of which are xylograph printings. Ben Nourse has linked the two Tibetan canonical recensions to the two versions of the text found at Dunhuang but has not yet published this work.

What Silk's study highlights, far better than Conze's rather chaotic work on the Sanskrit, is that the history of this text is complex. Even in this short text from the mid 7th Century, there are major variants before the end of the 8th Century, and dozens of variations have come down to us. Manuscripts both diverge and converge so that the usual stemma diagram of simple binary branches (based on Darwin's evolutionary tree?) is entirely inadequate to the task. The tree metaphor has strictly limited applicability (see my essay Evolution: Trees and Braids. 27.12.2013).

In the Dunhuang cache there is a T 215 transliterated in Tibetan script, a Tibetan translation of T 251, extended texts corresponding to TibA and TibB, as well as one or two hybrid texts that combine features of standard and extended texts.

Far more than in other traditions, Mahāyana Buddhists actively changed their sacred texts over time creating a plethora of versions. 

Chinese Texts

The extended text is seldom, if ever, referred to in China where the standard text (T 251) is used exclusively. However, five versions of the extended text are preserved in the Chinese canon. Chinese and Japanese commentaries are all on the standard version. At least two modern translations of these texts can be found on the internet (e.g. T 252 and T 253) but to the best of my knowledge, none has been published. 

Representing R1 we have:
  • T 253, translated by Bānruò 般若 (*Prajñā), with Lìyán 利言 et al.788 CE.
  • T 254, translated by Zhìhuì lún 智慧輪 (*Prajñācakra), 861 CE. 
  • T 255, translated from the Tibetan by Chos grub (aka Fǎchéng 法成), 856 CE.
  • T 257, translated by Dānapāla, 1005 CE.
And representing R2 we have:
  • T 252, translated by Fǎyuè 法月 (Skt. *Dharmacandra) with Lìyán 利言 ca. 741 CE
It is interesting that Lìyán 利言 was involved in both T 252 and T 253. I cannot find much information about him. We need a qualified Sinologist to look into the attributions and the dates associated with these texts. I cannot help but wonder if Prajñā (Bānruò 般若) and Prajñācakra (Zhìhuì lún 智慧輪) are definitely two different people. On the other hand, the two texts do have quite major differences.

T 255 was found in the Dunhuang cache.

TibA and TibB are both versions of R1. 

The structure of the Extended Text

The extended text adds material at the beginning and the end of the standard text. As we know, the standard text was considered by Chinese Buddhists to lack essential features of a Buddhist Sutra, specifically the opening phrase "thus have I heard", a statement of the place and occasion for the teaching, a statement of who was in the audience, the presence of the Buddha or the Buddha's invitation to speak; it does not conclude with an endorsement of the speaker (if it was not the Buddha), or an exhortation to disseminate the text, or the audience rejoicing in the teaching and committing to practice it.

The point is also made by Shāmén Yuánzhào 沙門圓照, in 794 CE, i.e. “However, [the Heart Sūtra] only has the main text and no nidāna or exhortation to disseminate” (但有正宗並 無序分及流通分。T 2157,  55.878c08). 

The extended versions add precisely these aspects.

It was previously thought that the longer text came first. This would be very unusual for an India text and is this not consistent with the India origins thesis. Neither Indian nor Chinese Buddhists would have removed the essential features of a sūtra from the extended version. Either could have added them. I'm not sure about Tibetans as I'm not so familiar with their textual practices, but they also seem unlikely to have removed these features. By far the most likely scenario, especially given the facts that have emerged in the last five years is that the extended version was created (probably twice) on the basis of the standard text.


In both recensions the standard opening paragraph is woven into the extended version, whereas the closing section (rejoicing) is simply tacked on after the dhāraṇī.

It appears that all the surviving versions contain a mistake that must go back to a common ancestor. Where in the standard text, Avalokiteśvara examined (vyavalokayati sma) the five branches of experience (pañca skandhāḥ) and saw (paśyati sma) that they are devoid of svabhāva (svabhāva-śunyan), the extended text inadvertently has vyavalokayati sma again for paśyati sma. The combination of "look/examine" (vyava√lok) and "see" (√paś) is not present in the Chinese, but is a canny addition to the Sanskrit text by whoever forged it. The combination of "look" and "see" works the same in Sanskrit as it does in English. The two languages also share the cognitive metaphor: TO SEE IS TO KNOW.  The difference is between seeking and finding, and it is lost in all the extant Sanskrit extended version and in both recensions the Tibetan as well.

By contrast the Chinese versions of the Extended text retain the Chinese syntax of the standard version, with just one verb, in which, Avalokiteśvara "clearly sees the five branches are absent" (zhàojiàn wǔyùn jiē kōng照見五蘊皆空). T 253 and 254 show further conflation with the standard text when they add to this "[he is] parted from all misery"  (lí zhū kǔ è 離諸苦厄。) which reflects the standard text's "transcending all miseries" (dù yī qiè kǔ è 度一切苦厄 。).

Another point of difference is the audience. Recension 1 has the Buddha on the Vulture Hill "with a great congregation of bhikṣus and a great congregation of bodhisatvas" (mahatā bhikṣusaṃghena sārdhaṃ mahatā ca bodhisatvasaṃghena). Bodhisatvasaṃgha is an unusual term. T 257 adds the detail that 1250 monks are present (與大苾芻眾千二百五十人俱).

The reason for believing that T 252 is a second Recension is that the details, while formulaic, are substantially different from the other versions in any language, while the other texts are all similar (though certainly not identical).  T 252 also specifies the size of the assembled congregations "together with a great bhikṣu congregation of  100,000 and 77,000 bodhisatva mahāsatvas in all" (與大比丘眾滿百千人,菩薩摩訶薩七萬七千人俱) and then names Avalokiteśvara, Mañjuśrī, and Maitreya as the leaders of the bodhisatvas. 


The closing passage of R1 is several paragraphs. Avakokiteśavara concludes, the Bhagavan emerges from his samādhi and praises Avalokiteśvara, Śārputra rejoices, then the whole world rejoices and everyone commits to practising the teaching. In R2 this is all condensed to a single generic paragraph, which matches the last part of R1.


I've said many times that the Heart Sutra has been neglected by academia but it has also been the victim of mistakes and misdirections that might have made it a less attractive prospect for academic study that its immense popularity amongst Buddhists implied.

If the standard text has been neglected, then outside of Tibetan religious circles the extended text has been abandoned. There are no studies of the extended text that I am aware of, other than those produced by the Dalai Lama and other Tibetan religious figures. Thus there is scope for continuing study. 

Many questions about the extended text remain open, including:

  • Why are there two recensions?
  • When was the Heart Sutra extended and by whom? 
  • Where the extensions added in Chinese, Sanskrit, or perhaps Tibetan?
  • Was the text known in India? 
  • Where do the dates and attributions of the Chinese texts come from?

I hope to be able to begin to address some of these questions in the coming weeks or months.


Related Posts with Thumbnails